Applied and Computational Mathematics 2019; 8(2): 37-43
http://www.sciencepublishinggroup.com/j/acm
doi: 10.11648/j.acm.20190802.13
ISSN: 2328-5605 (Print); ISSN: 2328-5613 (Online)
Association Rule Mining for Career Choices Among Fresh Graduates
Leibao Zhang1, Xiaowen Tan
1, Shuai Zhang
2, *, Wenyu Zhang
2 1School of Public Finance and Taxation, Zhejiang University of Finance and Economics, Hangzhou, China 2School of Information, Zhejiang University of Finance and Economics, Hangzhou, China
Email address:
*Corresponding author
To cite this article: Leibao Zhang, Xiaowen Tan, Shuai Zhang, Wenyu Zhang. Association Rule Mining for Career Choices Among Fresh Graduates. Applied and
Computational Mathematics. Vol. 8, No. 2, 2019, pp. 37-43. doi: 10.11648/j.acm.20190802.13
Received: May 12, 2019; Accepted: July 2, 2019; Published: July 19, 2019
Abstract: Nowadays, an increasing number of colleges have built information systems to manage masses of educational data,
but actually most data is in an idle state and fails to create any value. As an efficient data analysis method, association rule mining
can precisely make good use of these disordered data and extract useful but latent information from them. In this paper, an
example of 228 students, who graduated from the School of Information of Zhejiang University of Finance and Economics,
China in 2017, is taken to discover the association rules between their career choices and academic performance using Apriori
algorithm. The main purpose of this paper is to offer fresh graduates a reference to future career choices and help teachers guide
them in better career planning. The experimental results indicate that the courses students are good at largely affect their career
choices, although their overall career scope is not narrow.
Keywords: Data Mining, Association Rule Mining, Apriori Algorithm, Career Choice, Academic Performance
1. Introduction
Recently, with the constant enlargement of college
enrollment scale, higher education is gradually stepping into
the stage of massification. However, what follows is the sharp
increase in both the number of fresh graduates and the
employment pressure each year. How to enter a suitable
industry and further get an ideal job according to one’s own
academic background also becomes a puzzle for most fresh
graduates. Against this background, the research on graduate
employment attracts more attention [1-3], but few of them
make an approach to the relationship between the career
choices and individual academic performance through an
effective method.
Data mining is a process of extracting potentially useful
information from vast amounts of data, and it is also known as
educational data mining when it comes to educational-related
data [4-5]. Generally, data mining can be realized through the
methods of association rule mining, classification, prediction
and so on. Among them, association rule mining can be
applied to identify frequent patterns and interesting causal
relationships that satisfy the predefined minimum values of
parameters concerned [6]. In this paper, to explore the
relationship between the fresh graduates’ career choices and
their individual academic performance, an example of 228
students, who graduated from the School of Information of
Zhejiang University of Finance and Economics, China in 2017,
is taken as the research object. The Apriori algorithm based on
R programming is used to realize the association rule mining.
The rest of this paper is structured as follows. Section 2
summaries the current related works on association rule
mining and Apriori algorithm. Section 3 elaborates the
approaches to data preprocessing and association rule mining.
Section 4 presents the experimental results of association rule
mining. Conclusions and directions of future work are
discussed in Section 5.
2. Related Work
With the rapid development of data mining technology,
38 Leibao Zhang et al.: Association Rule Mining for Career Choices Among Fresh Graduates
association rule mining has become a quite active research
hotpot for its special advantages. Ho et al. [7] combined
association rule mining with fuzzy theory to help investors
decide the right trade time in financial market. Kumar and
Toshniwal [8] used k-means algorithm to group the road
accident locations based on accident frequency count, and
then applied association rule mining to discover the main
factors that affect the road accidents at these locations.
Mane and Ghorpade [9] predicted the students’ admission
tendency for each branch of engineering through
association rule mining with pattern growth method. Nahar
et al. [10] applied association rule mining to identify the
sick and healthy factors that cause heart disease for males
and females. To extract KANSEI knowledge, Shi et al. [11]
applied association rule mining and rough sets to explore
the relationship between form features and KANSEI
adjectives. Based on the multi-objective particle swarm
optimization algorithm, Tyagi and Bharadwaj [12] adopted
support and confidence as two objective functions and then
proposed a new approach to obtain direct and indirect
association rules with high quality. Wang et al. [13]
introduced a weighted fuzzy association rule mining
method to discover the alarm sequences, and the method
combines fuzzy sets, Apriori algorithm and alarm time
sequence metrics together.
The Apriori algorithm [14] popularized in association rule
mining has been widely used in various fields. D’Angelo et al.
[15] proposed a pervasive computing architecture based on a
trust model, in which Apriori algorithm is used to extract the
users’ behavioral patterns from their historical data during
network interactions. Deng et al. [16] built a causation
analysis model for traffic accidents, and the Apriori-Genetic
algorithm is used to mine the main influencing factors. To
make the mobile e-commerce shopping more convenient and
high-quality, Guo et al. [17] proposed an improved Apriori
algorithm according to the characteristics of the mobile
e-commerce recommendation system. Ilayaraja and
Meyyappan [18] used the Apriori algorithm to identify the
frequent diseases at given time periods according to a data of
patients from different areas. Combining the Apriori
algorithm with frequent subgraph tree, Nair and Thomas [19]
proposed a two-phase approach to discover the frequent
subgraphs in graph datasets. Prasanna and Ezhilmaran [20]
used an improved Apriori algorithm, together with a modified
Genetic algorithm, to help the investors select better stock
rules for their investment.
However, there exist few studies focusing on the
relationship between career choices and academic
performance through data mining, especially the method of
association rule mining. Thus, based on a data set supported
by Zhejiang University of Finance and Economics, this paper
explores the relationship between fresh graduates’ career
choices and their individual academic performance. To be
specific, the academic performance here is mainly depicted by
students’ course grades. The experimental results are
conducive to making students’ career objectives clearer and
improving their chance of being employed.
3. Data Preprocessing and Association
Rule Mining
This section is made up of two parts. One is a detailed
description of data preprocessing, and the other is the
elaboration of the Apriori algorithm used for association rule
mining.
3.1. Data Preprocessing
The raw data used in this paper includes the employment
and academic performance information of 228 fresh graduates
from Zhejiang University of Finance and Economics, who
graduated in 2017 and respectively majored in Information
Management and Information System, Computer Science and
Technology, E-Commerce and Software Engineering. Owing
to the existence of missing values and noises, it is essential to
preprocess the raw data for more desired experimental results.
For the employment information, three different types of
these students’ career choices are summarized based on the
national economy industry classification and the
characteristics of their majors, i.e., Work in Information
Industry, Work in Finance or Business, and Work in Other
Industries. For academic performance information, except for
the deletion of students’ private information like their names
and ID numbers, three main steps are taken as follows.
(1) The combination of similar attributes. The raw data of
academic performance is mainly consisted of students’ course
grades and the total number of the courses is 207. Among all
courses, some courses are rather similar, such as English (2),
English (3) and English (4). Taking these three courses as an
example, they are combined into a new attribute named
College English through averaging.
(2) The removal of unsuitable attributes and records.
Although the raw data contains the information of a lot of
courses, about 90% of the courses are taken by less than one
third of all 228 students. Thus, to make the experimental
results more accurate, these courses must firstly be removed.
Secondly, courses which gave the similar grades to all students
are removed, such as College Chinese, Military Training and
so on. And lastly, students with too many missing attribute
values are also removed.
(3) The transformation of course grades. According to the
Implementation Measures of Credit System for Undergraduate
Students of Zhejiang University of Finance and Economics, the
grade of one course can be given in either hundred-mark system
or grade point system, but grade point average (GPA) must be
employed to measure students’ overall quality. To satisfy the
data type requirement of association rule mining in R
programming, five levels listed in Table 1 are used to transform
the course grades and GPA. The NA values are labeled as
“Unselected” for those students who did not take the course.
Besides, although students must attend the make-up exams if
they fail the compulsory courses, they can take other courses
instead if they fail the optional courses. Thus, the label
“Unselected” is also given to the records of cancelling or
abandoning the exams belonging to optional courses.
Applied and Computational Mathematics 2019; 8(2): 37-43 39
Table 1. The transformation criteria of course grades and GPA.
Excellent Good Middle Pass Fail
Hundred-mark 100-90 89-80 79-70 69-60 <60
Grade point 5-4.5 4.4-3.5 3.4-2.5 2.4-1.5 0
After the data preprocessing expounded above, the data
used consists of the information of 205 students with their
career choice, Gender, GPA, 15 courses’ grades and so on. The
partial preprocessed data used for association rule mining is
shown in Table 2.
Table 2. The partial preprocessing results of data.
Career Choice Gender GPA College English M-Commerce E-Government Multimedia
Technology
Database
Curriculum Design
Work in Finance or Business Female Good Middle Excellent Middle Good Good
Work in Finance or Business Female Middle Good Good Unselected Middle Good
Work in Other Industries Female Good Good Middle Unselected Good Good
Work in Finance or Business Female Middle Middle Excellent Good Good Good
Work in Finance or Business Female Middle Middle Excellent Good Pass Good
Work in Other Industries Male Good Good Excellent Unselected Excellent Excellent
Work in Information Industry Male Middle Pass Middle Middle Middle Middle
Work in Other Industries Female Good Middle Unselected Excellent Good Middle
Work in Information Industry Male Good Middle Unselected Good Excellent Good
Work in Information Industry Male Good Middle Unselected Good Middle Middle
Work in Information Industry Female Good Good Unselected Middle Good Middle
Work in Other Industries Female Good Middle Unselected Excellent Good Good
Work in Information Industry Male Middle Middle Good Good Pass Middle
Work in Finance or Business Male Middle Good Pass Excellent Middle Excellent
Work in Finance or Business Female Middle Middle Excellent Good Fail Middle
Work in Information Industry Male Middle Middle Good Excellent Good Good
Work in Other Industries Male Middle Pass Good Good Good Middle
3.2. Association Rule Mining
In this paper, the R programming and Apriori algorithm are
used to discover the rules between the selected fresh graduates’
career choices and their academic performance. For the
Apriori algorithm, which centers mining the frequent itemsets
of Boolean association rules, its realized process can be
divided into the following two steps.
First, discover all the frequent itemsets whose support
values are not lower than the predefined threshold through
iterations.
Second, take advantage of frequent itemsets to generate the
rules which satisfy the predefined minimum confidence value.
Thus, two key parameters are included in this algorithm,
i.e., support and confidence. Assuming Z={Z1, Z2, …, Zm}
is a set of items, the traditional association rule is an
expression as X Y⇒ , in which X Z∈ , Y Z∈ , and
X Y∩ = ∅ . The support of each rule represents the
probability that X and Y simultaneously occur in item Z.
The confidence of each rule represents the probability that
Y occurs in the items where X has been included. After a
series of tests, the threshold of support and confidence
values are set as 0.02 and 0.75 respectively in this paper.
Furthermore, lift is also an important parameter which
should be considered. The lift of each rule represents the
ratio of its confidence and the probability that Y occurs.
When the value of lift is equal to 1, it means that there exists
no correlation between X and Y. Thus, the rule can be kept
only when its lift value is higher than 1. The computational
formulas of these three parameters are respectively shown
in equations 1 to 3, in which Number ( X Y∩ ) represents
the number of itemsets containing X and Y and Number (Z)
represents the number of all itemsets [21].
( ) ( ) ( )( )
Number X YSupport X Y P X Y
Number Z
∩⇒ = ∩ = (1)
( ) ( ) ( )( )
P X YConfidence X Y P Y X
P X
∩⇒ = = (2)
( ) ( )( )
( )( )
⇒⇒ = =
P Y XConfidence X YLift X Y
P Y P Y (3)
4. Statistical Analysis and Experimental
Results
In this section, a descriptive statistical analysis is firstly
given to show the basic information about the processed data
set. And then, the experiments results are elaborated to reveal
the potential relationship between the selected fresh graduates’
career choices and their academic performance.
4.1. Descriptive Statistical Analysis
In order to better know the used data, the descriptive
statistical analysis concerning the career choices and course
grades is conducted in this sub-section. As illustrated in Figure
1, students graduated with four different majors have a wide
career scope, but most students in each major still choose to
work in information industry, which is in accordance with
their general academic background. In terms of the remaining
two career choices, students majoring in Information
Management and Information System tend to work in other
industries like public administration, social work and so on.
On the other hand, students majoring in E-Commerce have a
40 Leibao Zhang et al.: Association Rule Mining for Career Choices Among Fresh Graduates
higher possibility to work in finance or business than other
majors, partly because they have more access to the courses
related to this area. since different learning experiences have
effect on students’ career choices to some extent [22].
Figure 1. The distribution of career choices for students with four different majors.
The overall level of students’ grades in 15 selected courses
is measured in quartiles and mean. These two methods usually
reflect the central tendency with different features. Table 3
gives the transformed results of the descriptive statistics and
the records with all grades kept for each course. The first
quartile shows that at least 25% students get poor grades
respectively in Database Principle and Analysis, Managerial
Operation Research, Mathematics and Accountancy. The
values of the median and mean are the same here and the
levels are either “Middle” or “Good”. As for the third quartile,
in all courses except College English, Database Principle and
Analysis, and Accountancy, the students get grades no lower
than the level “Good”. It is obvious that students are not good
at these three courses relative to the other ones. Thus,
improving their academic performance on these courses is of
great necessity, especially College English, because English
ability has been a key factor for students’ future development
now.
Table 3. The descriptive statistical results of all courses.
Course Number First Quartile Median Mean Third Quartile
College English 205 Middle Middle Middle Middle
M-Commerce 111 Middle Good Good Good
E-Government 122 Good Good Good Excellent
Multimedia Technology 175 Middle Good Good Good
Database Curriculum Design 205 Good Good Good Excellent
Database Principle and Analysis 125 Pass Middle Middle Middle
Business Data Analysis 88 Middle Good Middle Excellent
Enterprise Resource Planning 131 Good Good Good Good
Managerial Operation Research 111 Pass Middle Middle Good
Mathematics 125 Pass Middle Middle Good
Accountancy 158 Pass Middle Middle Middle
Marketing 125 Middle Good Good Good
Mobile Development Technology 140 Good Good Good Excellent
Management 124 Middle Good Good Good
Economic Principle 157 Middle Middle Middle Good
Computer Networks 140 Middle Good Good Good
4.2. Experimental Results
Figure 2 is the scatter plot that depicts the association rules
of students’ career choices. The value of support and
confidence are represented by the X axis and Y axis
respectively. The points in this figure represent the association
rules and the colors correspond to different values of lift. As
illustrated in Figure 2, the support values of most association
rules are lower than 0.1 and the confidence values of most
association rules are higher than 0.75. However, the lift values
are mostly between 1.5 and 4, which indicates that the students’
academic performance do have great effect on their career
choices. This finding is also supported by Uyar et al. [23], in
which academic performance is proven to influence whether
students choose to work in accounting.
Applied and Computational Mathematics 2019; 8(2): 37-43 41
Figure 2. Rules of students’ career choices.
Figure 3 is the connection graph of the association rules of
students’ career choices. The size of each circle represents the
support value of the rule, and the bigger the circle is, the
higher the support value is. Meanwhile, the color of each
circle represents the lift value of the rule, and the darker the
circle is, the higher the lift value is. Owing to the extremely
large quantity of the association rules, six rules for each career
choice are selected as examples. As illustrated in Figure 3,
most rules of students working in information industry have
higher support values but lower lift values. On the contrary,
the rules of students working in finance or business almost
have highest lift values but lowest support values. This may be
because most students choose to work in information industry
rather than finance or business.
Figure 3. Graph for rules of students’ career choices.
Table 4 gives the partial results of association rules between
students’ career choices and academic performance. Take
several rules listed as examples, rule 1 shows that male
students who get level “Good” in Enterprise Resource
Planning, Mobile Development Technology and Computer
Networks but level “Middle” in College English have a greater
possibility to work in information industries. Sheer interest in
one industry tends to have a positive effect on students’ career
choices. It has been proven by Edwards and Quinter [22] and
Lent et al. [24], that students’ interest is the main motives for
their career choices. Thus, the phenomenon revealed by rule 1
can be largely explained that those students are likely to have
more interest in information technology and more enthusiasm
to learn the related courses well, so that they can further work
in information industries more smoothly. Rule 3 shows that
female students who get level “Good” in Database
Curriculum Design and Mobile Development Technology tend
to work in finance or business. Employability skills play a key
role in determining students’ career choice. Having a good
command of the skills which are exactly sought by employers
offers most students bigger chance to be employed [25-26].
From this perspective, this phenomenon can be explained that
these two courses make female students equipped with the
employability skills to work in financial or business
organizations, such as the ability to develop the systems or
APPs that are often necessary by many of those organizations
against this era of big data. Rule 5 shows that students who get
level “Good” in GPA, Mobile Development Technology and
Multimedia Technology, and even level “Excellent” in
E-Government are more likely to work in other industries.
Seeking opportunities to show mastered skills and knowledge
is also an influential factor for students’ career choice [22].
The phenomenon revealed by this rule can largely be
explained that most of these courses have a wider applicability
in other industries like public administration or entertainment,
so the students with good GPA want to give full play to their
42 Leibao Zhang et al.: Association Rule Mining for Career Choices Among Fresh Graduates
advantages and further win a promising promotion prospect
for themselves if being employed in these industries. In
addition, getting a “Good” level in Mobile Development
Technology appears in each career choice, which can be
reflected in rules 1, 3, 5 and 6. This indicates that Mobile
Development Technology represents a relatively practical skill
for most students.
Table 4. The partial rules of students’ career choices.
Number Rules Support Confidence Lift
1
{Gender = Male, College English = Middle, Enterprise Resource Planning = Good, Mobile
Development Technology = Good, Computer Networks = Good} => {Career Choice = Work in
Information Industry}
0.024390 1 2.070707
2 {Managerial Operation Research = Pass, Accountancy = Middle, Economic Principle = Middle}
=> {Career Choice = Work in Information Industry} 0.048780 0.909091 1.882461
3 {Gender = Female, GPA = Middle, Database Curriculum Design = Good, Mobile Development
Technology = Good} => {Career Choice = Work in Finance or Business} 0.029268 0.857143 4.086379
4 {Gender = Male, M-Commerce = Good, Marketing = Good, Management = Good} => {Career
Choice = Work in Finance or Business} 0.024390 0.833333 3.972868
5 {GPA = Good, Mobile Development Technology = Good, E-Government = Excellent, Multimedia
Technology = Good} => {Career Choice = Work in Other Industries} 0.024390 1 3.253968
6 {Gender = Female, Enterprise Resource Planning = Good, Mobile Development Technology =
Good, Management = Good} => {Career Choice = Work in Other Industries} 0.039024 0.888889 2.892416
5. Conclusion
Nowadays, solving the problem of employment for fresh
graduates has become rather hot-debated. In this paper,
association rule mining and Apriori algorithm are applied to
explore the relationship between students’ career choices and
academic performance. Firstly, the raw information of 228
students graduated from the School of Information of
Zhejiang University of Finance and Economics in 2017 is
collected. Secondly, the raw data is preprocessed and
transformed to meet the requirements for the further study.
Thirdly, a data set containing 205 students’ career choices and
their academic performance is used to conduct the descriptive
statistical analysis and association rule mining. The
experimental results indicate that the selection among three
career choices is largely influenced by students’ grades in
different courses, but Mobile Development Technology
represents a relatively practical skill for most students owing
to its applicability to each career choice. The rules mined can
help students learn to understand their future career objectives
better based on their academic characteristics. On the other
hand, the rules can also help teachers to advise students on
their career planning.
However, there still exist some limitations that need to be
addressed in future work. For instance, the course information
is dispersive that only 15 courses are kept. The limited data set
influences the accuracy of experimental results. Moreover,
some other data mining technologies or heuristic algorithms
can be combined with association rule mining to improve the
efficiency and accuracy of the experimental results.
Acknowledgements
This work was supported by a grant from
Industry-university Cooperation and Cooperative Education
Project of the Ministry of Education, China
(No.201802170003, No.201701028082).
References
[1] Gao, L. (2015). Analysis of employment data mining for university student based on WEKA platform. Journal of Applied Science and Engineering Innovation, 2 (4), 130-133.
[2] Mishra, T., Kumar, D., & Gupta, S. (2016). Students’ employability prediction model through data mining. International Journal of Applied Engineering Research, 11 (4), 2275-2282.
[3] González-Romá, V., Gamboa, J. P., & Peiró, J. M. (2018). University graduates’ employability, employment status, and job quality. Journal of Career Development, 45 (2), 132-149.
[4] Chen, M. S., Han, J., & Yu, P. S. (1996). Data mining: An overview from a database perspective. IEEE Transactions on Knowledge and Data Engineering, 8 (6), 866-883.
[5] Romero, C., & Ventura, S. (2007). Educational data mining: A survey from 1995 to 2005. Expert Systems with Applications, 33 (1), 135-146.
[6] Solanki, S. K., & Patel, J. T. (2015). A survey on association rule mining. In Proceedings of the 5th International Conference on Advanced Computing & Communication Technologies, February 21-22, Haryana, India, pp. 212-216.
[7] Ho, G. T., Ip, W. H., Wu, C. H., & Tse, Y. K. (2012). Using a fuzzy association rule mining approach to identify the financial data association. Expert Systems with Applications, 39 (10), 9054-9063.
[8] Kumar, S., & Toshniwal, D. (2016). A data mining approach to characterize road accident locations. Journal of Modern Transportation, 24 (1), 62-72.
[9] Mane, R. V., & Ghorpade, V. R. (2016). Predicting student admission decisions by association rule mining with pattern growth approach. In Proceedings of International Conference on Electrical, Electronics, Communication, Computer and Optimization Techniques, December 9-10, Mysuru, India, pp. 202-207.
Applied and Computational Mathematics 2019; 8(2): 37-43 43
[10] Nahar, J., Imam, T., Tickle, K. S., & Chen, Y. P. P. (2013). Association rule mining to detect factors which contribute to heart disease in males and females. Expert Systems with Applications, 40 (4), 1086-1093.
[11] Shi, F., Sun, S., & Xu, J. (2012). Employing rough sets and association rule mining in KANSEI knowledge extraction. Information Sciences, 196, 118-128.
[12] Tyagi, S., & Bharadwaj, K. K. (2013). Enhancing collaborative filtering recommendations by utilizing multi-objective particle swarm optimization embedded association rule mining. Swarm and Evolutionary Computation, 13, 1-12.
[13] Wang, J., Li, H., Huang, J., & Su, C. (2016). Association rules mining based analysis of consequential alarm sequences in chemical processes. Journal of Loss Prevention in the Process Industries, 41, 178-185.
[14] Agrawal, R., Imieliński, T., & Swami, A. (1993). Mining association rules between sets of items in large databases. In Proceedings of the ACM SIGMOD International Conference on Management of Data, May 26-28, Washington D. C., USA, pp. 207-216.
[15] D’Angelo, G., Rampone, S., & Palmieri, F. (2017). Developing a trust model for pervasive computing based on Apriori association rules learning and Bayesian classification. Soft Computing, 21 (21), 6297-6315.
[16] Deng, X., Zeng, D., & Shen, H. (2018). Causation analysis model: Based on AHP and hybrid Apriori-Genetic algorithm. Journal of Intelligent & Fuzzy Systems, 35 (1), 767-778.
[17] Guo, Y., Wang, M., & Li, X. (2017). Application of an improved Apriori algorithm in a mobile e-commerce recommendation system. Industrial Management & Data Systems, 117 (2), 287-303.
[18] Ilayaraja, M., & Meyyappan, T. (2013). Mining medical data to identify frequent diseases using Apriori algorithm. In Proceedings of International Conference on Pattern Recognition, Informatics and Mobile Engineering, February 21-22, Salem, India, pp. 194-199.
[19] Nair, J. J., & Thomas, S. (2017). Improvised Apriori with frequent subgraph tree for extracting frequent subgraphs. Journal of Intelligent & Fuzzy Systems, 32 (4), 3209-3219.
[20] Prasanna, S., & Ezhilmaran, D. (2016). Association rule mining using enhanced Apriori with modified GA for stock prediction. International Journal of Data Mining, Modelling and Management, 8 (2), 195-207.
[21] Agrawal, R., Srikant, R. (1994). Fast algorithms for mining association rules. In Proceedings of the 20th VLDB Conference, September 12-15, Santiago, Chile, pp. 487-499.
[22] Edwards, K., & Quinter, M. (2011). Factors influencing students career choices among secondary school students in Kisumu Municipality, Kenya. Journal of Emerging Trends in Educational Research and Policy Studies, 2 (2), 81-87.
[23] Uyar, A., Güngörmüs, A. H., & Kuzey, C. (2011). Factors affecting students’ career choice in accounting: The case of a Turkish university. American Journal of Business Education, 4 (10), 29-38.
[24] Lent, R. W., Brown, S. D., Talleyrand, R., McPartland, E. B., Davis, T., Chopra, S. B., Alexander, M. S., Suthakaran, V., & Chai, C. M. (2002). Career choice barriers, supports, and coping strategies: College students’ experiences. Journal of Vocational Behavior, 60 (1), 61-72.
[25] Wye, C. K., & Lim, Y. M. (2009). Perception differential between employers and undergraduates on the importance of employability skills. International Education Studies, 2 (1), 95-105.
[26] Bridgstock, R. (2009). The graduate attributes we’ve overlooked: Enhancing graduate employability through career management skills. Higher Education Research & Development, 28 (1), 31-44.